Explainer: Why Taiwan is down but not out in the global AI race
Artificial intelligence (AI) is transforming our businesses, industries and the global economy at large. As other countries swarm to seize emerging opportunities brought by AI, Taiwan, the island situated amongst two AI front-runners—China and Japan—seems to have trouble rising above the clamor.
The high-tech island is home to some of the world’s leading hardware and electronics companies, including TSMC, Foxconn, Quanta, and Asus. But despite its tech prowess, Taiwan’s AI development is still at a nascent stage.
Tech talent exodus
As with the rest of the world, much buzz has been generated in Taiwan by the AI boom. However, this has been accompanied—some may say inevitably—by rising concerns over the brain drain in Taiwan’s tech industry.
But it wouldn’t be fair to say that Taiwan is unable to foster talents. In fact, Taiwan has a highly educated workforce abundant with tech talents. Students from Taiwan demonstrate high performance in mathematics and science, the 4th highest in the world according to OECD, and over 25% of all its university degrees are in engineering. Kai-Fu Lee, AI expert and founder of Sinovation Ventures, and Aja Huang, AlphaGo’s lead programmer, are among the many Taiwanese who are highly respected in the field.
Taiwan’s tech industry has long been dominated by big semiconductor and IT manufacturing companies that have the resources and money to offer higher salaries and better opportunities to attract talents than software companies. Fresh graduates in Taiwan often have to face the dilemma of choosing between the less lucrative positions in the software industry and the higher-paid jobs in the traditional hardware industry. The established hardware industry inadvertently undermines the development and innovation of software.
Much for the same reason, Taiwan hasn’t been able to stop the ongoing exodus of top AI engineers and technical talents from leaving for large international enterprises overseas. In Taiwan, the starting salary for fresh out of college software engineers is around $15K, and $16.8K for those with a master’s degree according to 1111 Job Bank, a local website for human resources—meager considering software engineers in the US on average earn around $100K a year.
“Currently, options are limited for the many outstanding Taiwanese graduates that are interested in AI. Most of them have to go abroad to join Facebook or Google for a proper career. If they join Taiwanese corporations such as TSMC or MediaTek, they won’t have enough opportunities to work in the field of AI,” Ethan Tu, the famed creator of Taiwan’s popular online bulletin board system PTT, said in April when he announced the launch of Taiwan AI Lab.
Most recently, Tu was made Regional Director of Microsoft Asia-Pacific Research and Development for AI but decided to return to his roots to help build an AI ecosystem in Taiwan. The Lab, aiming to push forward new ideas and product innovation in AI, reportedly offers the highest salary a software engineer can make in Taiwan.
Waking up to the AI revolution
Holding on to its legacy in the hardware industry, Taiwan seems to have sat quietly on the sidelines through the significant changes happening in software, the internet, and social networking in the early 2000s. Still boxed in the old PC-era way of thinking, Taiwan still lags behind other developed countries in providing an enabling financial and regulatory environment to cultivate the development of AI and deep-learning.
So, what is the island doing counter these setbacks?
In August, the Ministry of Science and Technology (MOST) laid out a $534.5 million development plan for AI over the course of the next four to five years. One of the initiatives, Project Moon Shot, is a $132 million project to assist Taiwan’s semiconductor industry in developing cutting-edge AI technologies and edge products, and cultivating a talent pool specializing in AI. Resourceful tech companies including Microsoft Taiwan and NVIDIA are offering support to the government in carrying out the development plan.
In November, the national academy, Academia Sinica, announced the opening of Taiwan’s first AI research school—Taiwan AI Academy—which is set to begin its first program in January 2018.
Looking for an edge in AI
Taiwan doesn’t have a billion people to generate the amount of data needed to train AI systems, nor does it have the size of talent pool comparable to that of China and US, but does it really need to depend on those conditions to get a foothold in AI?
In an interview with Manager Today, Kai-Fu Lee implied that the area of opportunity for Taiwan is not necessarily in AI’s core technologies—namely machine learning and deep learning—but the bigger opportunities lie in hardware and industrial AI applications.
1. AI boom presents vast opportunities for the semiconductor industry
The semiconductor industry is at the heart of the development of AI applications. AI chips are used in data centers to train systems to analyze and find patterns in volumes of data that are compiled. That is why semiconductor companies are racing to develop AI chips that could boost deep-earning training times.
Home to two of the world’s largest semiconductor companies—Taiwan Semiconductor Manufacturing Company (TSMC) and United Microelectronics Corporation (UMC)—Taiwan holds the world-leading status in the supply chain of semiconductors. The industry is accounted for 28% of the Taiwan’s exports in 2016, which means the AI chip boom is a golden opportunity.
2. Edge AI
Many experts believe the best opportunity that matches Taiwan’s strength lies on the edge, not cloud.
Smart home appliances, smart surveillance, smartphone, robotics, drones, and IoT devices, there are many uses cases where it is essential to run deep-learning algorithms at the edge to avoid latency in receiving and processing data.
Since the endpoint devices take very specific AI-chips and low-power technologies, tech companies are still exploring concepts and developing technologies that will bring AI closer to the edge—this is where Taiwan’s competitive advantage lies.
3. Surveillance cameras and sensors
Video surveillance cameras and sensors are used in factories and retail environments to gather data that can be analyzed to extract valuable insights. Taiwan already has many outstanding optical lens companies like Largan Precision, which is a supplier of camera lens modules for smartphones, tablet computers, and digital cameras, among other devices that can benefit from the growing demand.
4. Industrial AI applications—unmanned transportation and robotics
Hardware manufacturing will still play an important role in the AI era. The demand for components and parts for robotics and unmanned vehicles are sure to arise in the future. Smart manufacturing system and equipment is also an area Taiwan could tap into. For example, Foxconn, Apple’s supplier and the world’s largest contract electronics manufacturer, is already bringing AI to its factories. Industrial robots, built almost entirely in-house, are already deployed in the company’s smartphone assembly lines.
When AI and automation become prevalent, “the robotics hardware sector [in Taiwan] can benefit the most,” said Albert Chang, managing partner of McKinsey’s Taiwan office.
Developing a software mentality
While leveraging its competitive edge is essential, it doesn’t mean that Taiwan can overlook the development in software industry. If Taiwan persists in being the manufacturing yard for other technology advanced countries and sits out on yet another major software revolution, it will eventually fall behind in the AI race. After all, AI’s core technology is software such as big data, deep-leaning, and algorithms.
Ultimately the challenges will be in creating an environment that enables a thriving software culture and a burgeoning talent pool.
Taiwan AI startups pushing the new frontiers and bringing new innovations
There is an increasing number of AI startups surfacing in Taiwan. Here are a few that have caught our eye:
Appier aims to provide AI platforms to help enterprises solve their challenging business problems. This year, the company has been named in CB Insights’ “100 Companies 2018” for the second time.
Umbo, one of the rising stars in Taiwan’s AI community, are using AI to get a more precise read from security cameras. Their AI-powered technology is capable of detecting risks and abnormalities by scanning images from multiple security cameras.
DT42 is a deep-learning startup that aims to make AI easier and more affordable for businesses to deploy to edge devices.
AI is taking the world by storm, transforming all industries and sectors. Taiwan maybe standing on a shifting ground now, but there are reasons to be optimistic: the society’s passion around mathematics and science, the quality of its workforce, and its legacy in IT manufacturing and semiconductor. If Taiwan knows how to leverage its edge, it might have what it takes to rise to the AI forefront.